बग / वैशिष्ट्य विनंती करीता अहवाल द्या
Free.ai Platform

One API for every AI modality

Chat, vision, image generation and editing, speech to text, text to speech, OCR, translation, video, music, 3D and embeddings. One key, one base URL, one balance. Built for products and site operators who would otherwise be integrating five vendors.

See platform plans Talk to us about volume
48 self-hosted models 14 modalities on one key 476 external models available 477 tools we built on it

Most AI APIs sell you one modality

If your product needs a chat assistant, you buy language tokens. Then marketing wants images, so that is a second vendor. Support wants call transcription, a third. Finance wants invoice OCR, a fourth. Localisation wants translation, a fifth. Five contracts, five keys, five sets of rate limits and five invoices to reconcile, for one product.

Free.ai runs all of it. The models are self-hosted on our own hardware and reached through one base URL with one key, and the ones we do not host are routed to leading external providers behind the same key. You integrate once and add a modality by changing a path, not by starting a procurement process.

What you can call

Same key, same billing balance, same account. Model counts are self-hosted engines behind each endpoint.

Modality Endpoint What it does
Chat and reasoning
8 self-hosted
 POST /v1/chat/completions Drop-in chat-completions endpoint with streaming, function calling and long context. Point an existing client at our base URL and it works.
Vision
3 self-hosted
 POST /v1/image/describe/ Read a photo, screenshot, chart or receipt. Send an image inside a chat message or call the describe endpoint directly.
Image generation
4 self-hosted
 POST /v1/image/generate/ Text to image at native resolution, plus pose, depth and edge conditioning for art direction you can reproduce.
Image editing
6 self-hosted
 POST /v1/image/edit/ Instruction editing, background removal and upscaling on the same key. No second vendor for the retouch step.
Text to speech
12 self-hosted
 POST /v1/tts/ Batch or streaming speech, multilingual voices, and voice cloning from a short reference clip.
Speech to text
2 self-hosted
 POST /v1/stt/ Transcription with word timings and language auto-detection, for calls, meetings, podcasts and video.
OCR and document AI
4 self-hosted
 POST /v1/ocr/ Scans and PDFs to clean text or Markdown, with layout and table structure preserved.
Translation
1 self-hosted
 POST /v1/translate/ Machine translation across a very wide language set, plus document and subtitle translation on the same balance.
Video generation
3 self-hosted
 POST /v1/video/generate/ Text to video and image to video, plus trim, compress, subtitle burn-in and format conversion utilities.
Music and audio
2 self-hosted
 POST /v1/music/generate/ Generate tracks from a prompt, or split an existing mix into vocal and instrument stems.
3D
2 self-hosted
 POST /v1/3d/generate/ Single image to a textured 3D mesh, for product spins, game props and AR previews.
Embeddings
1 self-hosted
 POST /v1/embeddings/ Multilingual dense vectors for search, dedupe, clustering and retrieval augmented generation.
Web search and research POST /v1/search/ A live web index with cited answers, plus a multi-step research endpoint that plans sub-questions and returns a sourced report.
Documents and media utilities POST /v1/pdf/convert/ PDF merge, split, compress, sign and convert, plus audio and video processing. Unglamorous, and the reason half an integration exists.

One key, one base URL

The chat endpoint is chat-completions compatible, so most existing clients work by changing two settings. Everything else is a plain POST against the same host with the same bearer token.

Generate a key under Account, API. Full reference and language examples live on the API and developer pages.

API reference Developer dashboard 
# chat, streaming, chat-completions compatible
curl https://api.free.ai/v1/chat/completions \
  -H "Authorization: Bearer sk-free-..." \
  -d '{"model":"qwen3-8b","messages":[{"role":"user","content":"hi"}]}'

# same key, different modality
curl https://api.free.ai/v1/image/generate/ \
  -H "Authorization: Bearer sk-free-..." \
  -d '{"prompt":"studio photo of a ceramic mug","model":"sdxl"}'

curl https://api.free.ai/v1/stt/ \
  -H "Authorization: Bearer sk-free-..." \
  -F file=@call.mp3

curl https://api.free.ai/v1/ocr/ \
  -H "Authorization: Bearer sk-free-..." \
  -F image=@invoice.png

Why this is credible

It already runs our own products

Free.ai's public tool catalogue, our Android and iOS apps and our sibling sites all call this exact API. There is no separate enterprise stack, and no endpoint you get that we do not depend on ourselves every day.

The capacity already exists

Self-hosted inference runs on our NVIDIA H200 in New York, provisioned and serving traffic right now. Onboarding you is issuing an API key, not waiting on hardware. Live status is published at /status/.

Open weights, no lock-in

The self-hosted models are open-weight and named openly, so you always know what produced your output and you can reproduce it elsewhere if you ever want to. Outputs are yours, including for commercial use.

Platform plans

One token balance across every modality. Unlimited team seats on every tier. Cancel anytime.

Business

$49/mo

6.25M tokens per month, across every modality

For a product with steady but modest AI usage: a chat assistant, generated thumbnails, transcription on demand.

Start plan

Team

$99/mo

15M tokens per month, across every modality

For an agency or a production content pipeline, where several people generate every day against one budget.

Start plan
Most B2B accounts

Enterprise

$499/mo

100M tokens per month, across every modality

For a product serving AI features to your own users, where volume is continuous rather than bursty.

Start plan

Every tier above includes the same thing. They differ only in how many tokens you get each month.

  • Every modality on this page, on one key and one balance
  • 1,000,000 API requests per month
  • Unlimited team seats drawing on the same balance
  • A separate API key per developer, all billed centrally
  • Priority over free traffic in the generation queue
  • Invoicing and purchase orders on request, and cancel anytime
If video is your main use case, talk to us before you buy. Video is by far the heaviest workload we run: a short clip costs roughly 5,000 tokens against about 100 for a chat turn, so a video-led product burns a monthly allowance very differently from a text-led one. We would rather size it with you up front than have you run out in week two.

Talk to us about your volume

Custom volume, invoicing, purchase orders, net terms, or a modality mix that does not fit the plans above. Tell us what you are building and a person will reply, usually within one business day.

Or write straight to hello@free.ai

Buying for a team of people rather than a product? See Free.ai for Teams. Need a deployment shape other than our cloud? See Enterprise.

वारंवार विचारले जाणारे प्रश्न

Every modality on this page, billed from one balance: chat, vision, image generation and editing, text to speech, speech to text, OCR, translation, video, music, 3D, embeddings, web search and file processing. Same key, same base URL, same billing.

Most AI APIs sell language tokens. If your product also needs images, speech, transcription or OCR, you end up integrating four vendors, holding four sets of credentials and reconciling four invoices. Free.ai runs all of those itself, so it is one integration and one bill.

The chat endpoint is chat-completions compatible, so most existing SDKs and clients work by changing the base URL and the key. The other modalities are plain JSON or multipart POSTs, documented on /api/ and /developer/.

Self-hosted models run on our NVIDIA H200 in New York. That capacity is already provisioned and already serving traffic, so onboarding a new account is issuing an API key, not a procurement cycle. Models we do not host are routed to leading external providers and priced separately.

We are. Free.ai's own tool catalogue, our Android and iOS apps and our sibling sites all run on this exact API. There is no separate enterprise stack held back for paying customers, and no feature you get that we do not use ourselves.

One token balance across every modality. A plan tops your balance up each month, purchased tokens do not expire, and usage draws from the same wallet whether the call came from the web app, the mobile app or your server. No per-modality contracts and no per-seat fee.

Request caps are tied to your plan: 10,000 requests a month on Basic and Starter, 100,000 on Pro, and 1,000,000 on Business ($49/mo) and above. Tokens are metered separately from request count, so a heavy month is a balance question, not a hard stop.

Yes. Create a free account, generate a key under /account/?tab=api and run against the free daily allowance. It is the same endpoints and the same models as a paid key, just a smaller daily budget and a lower request cap.

Talk to us before you buy. Video is by far the heaviest workload on the box, roughly 5,000 tokens for a short clip against about 100 for a chat turn, so a video-led product burns a plan allowance very differently from a text-led one. We would rather size it correctly up front than watch you run out in week two.

Standard plans bill by card and email a receipt for every charge. For an invoice, a purchase order or net terms, use the form on this page or write to hello@free.ai and we will arrange it manually.

Being straight with you: Free.ai is a small operation running a single GPU host, and we do not publish a contractual uptime SLA we cannot yet stand behind. Live status is at /status/. Support is email, usually answered the same day. If you need a signed availability commitment, say so in the form and we will tell you honestly whether we can meet it.

It lands in our support queue and a person reads it, usually within one business day. Include your expected monthly volume and which modalities you need and the first reply will contain real numbers instead of a discovery call request.

Build on one API instead of five

Every modality, one key, one balance. Start on a plan today or tell us your volume and we will size it with you.

Start plan Talk to us

Free.ai ला आवडते?

या पानाला गुण द्या